Main sources of bias in clinical studies
| Type of bias | Description |
| Selection bias | Missing or inadequate randomisation: participant characteristics, including potential confounders, may differ between the experimental arms. |
| Performance bias | Missing or inadequate blinding of outcome assessor: a physician may be induced to detect improvements in treated individuals. |
| Detection bias | Missing or inadequate blinding of participants: an individual may be induced to perceive health improvements if they know they are being treated. |
| Attrition bias | Too many withdrawals, or unbalanced dropout rates among the arms: loss in statistical power, misleading outcome interpretation. |
| Reporting bias | Selective outcome reporting: in general, authors may be induced to select the findings to report (for example, due to presence of a funding institution). |